We study the magnetic properties of Ho thin films with different crystallinity (either epitaxial or non-epitaxial) and investigate their proximity effects with Nb thin films. Magnetic measurements show that epitaxial Ho has large anisotropy in two different crystal directions in contrast to non-epitaxial Ho. Transport measurements show that the superconducting transition temperature (Tc) of Nb thin films can be significantly suppressed at zero field by epitaxial Ho compared with non-epitaxial Ho. We also demonstrate a direct control over Tc by changing the magnetic states of the epitaxial Ho layer, and attribute the strong proximity effects to exchange interaction.

Superconducting spintronics is an emerging field with great promise for information processing.1 The superconducting equivalent of a conventional spintronic spin-valve exploits the exchange-splitting induced by proximity-coupling between a superconductor (S) and two or more ferromagnet (F) layers such that the proximity-induced suppression of the critical temperature (Tc) can be controlled by the relative magnetization directions of the F layers.2,3 Experimental confirmation of these effects has been obtained in both F/S/F4–8 and F/F/S9–12 geometries; however, the maximum change in TcTc) so far reported is ∼40 mK. In part, the small ΔTc is due to the short-range of the magnetic exchange coupling and the thickness of conventional F layers needed to maintain ferromagnetism. In this letter we show that a magnetic-field induced transition between F and antiferromagnetic (AF) phases in a single layer of the rare-earth ferromagnet Ho results in zero-field ΔTc of over 100 mK in a Nb layer.

In single-crystals Ho becomes AF phase at 133 K with a basal plane helical magnetic structure;13–17 this then transforms into a conical magnetic structure around 20 K with the magnetic moment tilting 10° away from the basal plane.15,16,18 Ho can be transformed into F phase by applying a large magnetic field parallel to the basal plane.14,17 We show below that our films behave in a similar way.

All films were grown by DC magnetron sputtering on a-plane (110) sapphire substrates. With liquid nitrogen cooling, a base pressure around 10−8 Pa was obtained with the water partial pressure below 10−9 Pa. Epitaxial Nb (110) films (tNb 15–30 nm) were grown at a substrate temperature (TS) of 880 °C, and a deposition rate (rD) of 0.035 nm s−1. Epitaxial Ho (002) films (12 nm) were grown on the Nb18,19 without breaking vacuum at TS = 650 °C, or at room temperature for non-epitaxial Ho (or Y) growth, with rD = 0.058 nm s−1. The films were capped with one 8 nm (non-superconducting) Nb layer to prevent oxidation. In-plane (IP) and out-of-plane (OOP) magnetic measurements were performed at 10 K using a vibrating sample magnetometer and a SQUID magnetometer, respectively. Transport measurements were carried out in a four-point geometry.

X-ray diffraction (XRD) was used for structural characterization of thick Nb (250 nm)/Ho (70 nm) epitaxial bilayers. The results shown in Fig. 1 indicate that the Nb and Ho films are both epitaxial. Non-epitaxial Ho is predominantly textured in the (002) direction but with other orientations present.20 

FIG. 1.

(a) The XRD 2θ-ω scan of an epitaxial Nb (250 nm)/Ho (70 nm) bilayer on a-plane sapphire. (b) The XRD Phi scans of a-plane sapphire (104), Nb (200), and Ho (101) peaks.

FIG. 1.

(a) The XRD 2θ-ω scan of an epitaxial Nb (250 nm)/Ho (70 nm) bilayer on a-plane sapphire. (b) The XRD Phi scans of a-plane sapphire (104), Nb (200), and Ho (101) peaks.

Close modal

Typical IP magnetic moment vs. magnetic field (M(H)) loops at 10 K are shown in Fig. 2(a). The non-epitaxial films approach saturation at 4 T with a volume magnetization of 1700 emu cm−3 and a remanence (Mr) close to zero which are comparable with previously reported values.20 

FIG. 2.

(a) IP M(H) loops of Nb/Ho bilayers. The magnetic field was swept from +4 T to −4 T, then from −4 T back to +4 T. (b) M(H) loops for the epitaxial Nb/Ho bilayer in a field range of 0.2 T and 0.5 T. (c) and (d) M(H) of Nb/Ho (epitaxial, non-epitaxial) bilayers for successively larger fields. (e) OOP M(H) loops of Nb/Ho bilayers. Inset top left: OOP magnetic moment vs. magnetic field of Nb/Ho bilayers, in which Ho is epitaxial. The magnetic field was first increased from 0 T to 0.1 T and removed. This sequence was continued with increasing fields. Inset bottom right: OOP magnetic moment vs. magnetic field of Nb/Ho bilayers, in which Ho is non-epitaxial. The field sequence is the same as epitaxial Ho. All measurements were performed at 10 K.

FIG. 2.

(a) IP M(H) loops of Nb/Ho bilayers. The magnetic field was swept from +4 T to −4 T, then from −4 T back to +4 T. (b) M(H) loops for the epitaxial Nb/Ho bilayer in a field range of 0.2 T and 0.5 T. (c) and (d) M(H) of Nb/Ho (epitaxial, non-epitaxial) bilayers for successively larger fields. (e) OOP M(H) loops of Nb/Ho bilayers. Inset top left: OOP magnetic moment vs. magnetic field of Nb/Ho bilayers, in which Ho is epitaxial. The magnetic field was first increased from 0 T to 0.1 T and removed. This sequence was continued with increasing fields. Inset bottom right: OOP magnetic moment vs. magnetic field of Nb/Ho bilayers, in which Ho is non-epitaxial. The field sequence is the same as epitaxial Ho. All measurements were performed at 10 K.

Close modal

Figure 2(b) shows IP low-field M(H) loops for an epitaxial sample: the 0.2 T loop is typical of the AF phase with an initial susceptibility similar to that of the non-epitaxial film. For the 0.5 T loop, the initial magnetization curve demonstrates a transition to square M(H) loop which is stable over subsequent field cycles. Unlike the virgin curve for a typical F, the majority part of the initial magnetization curve lies outside the loop demonstrating a definite phase transition to the F state.15 The high field M(H) loop for the epitaxial film reaches approximately 3000 emu cm−3—close to the theoretical saturation value of 10.34 μB.21 

To explore the AF-F phase transition of epitaxial films further we also measured Mr following successively higher applied fields (Happ) (Figs. 2(c) and 2(d)). For epitaxial Ho, Mr is negligible when Happ < 0.2 T, but increases for Happ ≥ 0.3 T and saturates at 0.6 T with a value of ∼2600 emu cm−3. Mr of non-epitaxial Ho remains low for all Happ.

OOP M(H) loops at 10 K are shown in Fig. 2(e): the non-epitaxial Ho shows a similar loop shape to the IP case but the magnetization is lower presumably because of the significant shape anisotropy and the strong-basal plane anisotropy of Ho.14 

For the epitaxial sample, a comparison of the IP and OOP magnetization at 4 T suggests a tilt angle of 15° away from the basal plane, comparable to single-crystal value.15 The absence of a significant Mr suggests that the magnetization may relax back to the basal plane after the field is removed.

The proximity effect between Ho and Nb was investigated by determining the bilayer Tc. Figure 3 summarizes measurements in the virgin state (i.e., before applying magnetic field) where Tc is defined from resistance vs. temperature R(T) curves as the temperature where the resistance drops to 50% of the residual resistance (Rr). A remarkable feature is that Tc is drastically suppressed by epitaxial Ho compared with non-epitaxial Ho. This suppression becomes larger as tNb is decreased, and is >3 K when tNb is 15 nm.

FIG. 3.

Tc vs. Nb thickness for bilayers of epitaxial Nb/non-epitaxial Y, epitaxial Nb/non-epitaxial Ho, epitaxial Nb/Ho. Inset: R(T) curves of an epitaxial Nb/Ho bilayer before and after Ho etching. The Ho layer was removed by first etching the Nb capping layer with CF4 plasma, and then etching the Ho layer with dilute hydrochloric acid which does not etch the underlying Nb layer. All measurements were performed at zero field.

FIG. 3.

Tc vs. Nb thickness for bilayers of epitaxial Nb/non-epitaxial Y, epitaxial Nb/non-epitaxial Ho, epitaxial Nb/Ho. Inset: R(T) curves of an epitaxial Nb/Ho bilayer before and after Ho etching. The Ho layer was removed by first etching the Nb capping layer with CF4 plasma, and then etching the Ho layer with dilute hydrochloric acid which does not etch the underlying Nb layer. All measurements were performed at zero field.

Close modal

To explore the effect of the epitaxial Ho AF-F phase transition on the proximity effect, we measured Tc following successively increasing set fields Hset (field applied and then removed). Figures 4(a) and 4(b) show that Mr and Tc exhibit a clear reciprocal correlation. No significant Tc shift could be detected for non-epitaxial samples (Fig. 4(c)) nor OOP fields (Figs. 4(d) and 4(e)). This behavior is reproducible for different tNb; here we only show representative data in a similar range.

FIG. 4.

The Mr of Ho and Tc of Nb/Ho bilayers measured with increasing Hset. (a) epi Nb (24 nm)/epi Ho (12 nm) with IP Hset. Inset left: IP AF state; right: IP F state. (b) epi Nb (30 nm)/epi Ho (12 nm) with IP Hset. (c) epi Nb (15 nm)/non-epi Ho (12 nm) with IP Hset. (d) epi Nb (30 nm)/epi Ho (12 nm) with OOP Hset. (e) epi Nb (15 nm)/non-epi Ho (12 nm) with OOP Hset.

FIG. 4.

The Mr of Ho and Tc of Nb/Ho bilayers measured with increasing Hset. (a) epi Nb (24 nm)/epi Ho (12 nm) with IP Hset. Inset left: IP AF state; right: IP F state. (b) epi Nb (30 nm)/epi Ho (12 nm) with IP Hset. (c) epi Nb (15 nm)/non-epi Ho (12 nm) with IP Hset. (d) epi Nb (30 nm)/epi Ho (12 nm) with OOP Hset. (e) epi Nb (15 nm)/non-epi Ho (12 nm) with OOP Hset.

Close modal

Before discussing more fundamental origins of the strong Tc suppression for epitaxial bilayers it is important to exclude trivial explanations. One such is the diffusion of Ho atoms as magnetic impurities into the Nb during growth. We compared the Tc for one sample before and after selectively etching the Ho from the surface as shown in the inset to Fig. 3. After Ho removal Tc increases from 6.1 K to 8.6 K (a value typical for Nb films) and therefore the Tc suppression is caused by a proximity effect rather than contamination.

A second mundane explanation could be the injection of OOP magnetic flux into Nb thin films arising from any OOP moment of the Ho. If this arose from the canted moment of the conical phase, then this should be larger for the epitaxial films and hence consistent with the lower Tc for the epitaxial samples; however, previous measurements of similar epitaxial films (albeit with thicker Nb base layers) concluded that the spiral period never became commensurate with the c-axis spacing,18 a condition believed to drive the tilt of the moments. Consequently, the zero-field c-axis moment may be close to zero, consistent with our OOP magnetization measurements reported here. Even if present, the canting angle should not be affected by the AF-F phase transition17 which is inconsistent with the increased Tc suppression in the F state.

We now consider the potential effect of the exchange coupling on Tc. The effect of non-epitaxial Ho on the epitaxial Nb is relatively weak. From the measured resistivity of one 30 nm epitaxial Nb film (3.6 μΩ cm), we can calculate the electron mean free path (lNb) and coherence length (ξNb) as 17 and 32 nm, respectively (assuming the Fermi velocity as 1.37 × 106 m/s20). From our XRD data (not shown) the grain size of non-epitaxial Ho (

$d_{\rm Ho_{non-epi}}$
d Ho non epi ⁠) is estimated to be about 4 nm. Since ξNbepi
$d_{\rm Ho_{non-epi}}$
d Ho non epi
, the mean exchange field experienced by Cooper pairs is averaged out; if the mean exchange field experienced by the Cooper pair is zero, then Ho behaves as if it is non-magnetic. This is confirmed by comparison with non-epitaxial Y also shown in Fig. 3; the small deviation for the lowest tNb may arise because lNb and ξNb will decrease with film thickness and the pairs may experience a net exchange field. To illustrate this point further, previously published data on Nb/Ho bilayers in which neither layer was epitaxial22 with comparable tNb show a Tc between 5 and 6 K for 12 nm Ho, i.e., lower than the samples just considered, but higher than the all-epitaxial samples; ξNbnon-epi ∼10 nm22 and so not much larger than
$d_{\rm Ho_{non-epi}}$
d Ho non epi
. Thus the averaging of the exchange field acting on Cooper pairs is reduced and Tc is significantly suppressed.

In contrast, for all-epitaxial bilayers the magnetic properties are invariant in the plane of the Nb/Ho interface and therefore, provided the Ho domain size is larger than ξNb, uniform over the lengthscale of ξNb. Although Ho has a helical magnetic structure along the c-axis, the magnetic moment in the basal plane is aligned ferromagnetically, which means that the alignment at the interface is also ferromagnetic. The Ho moment rotates in successive layers; nevertheless provided the lengthscale over which pairs sample the magnetic properties is comparable to the spiral wavelength (λ) the mean exchange field should remain considerable so as to impose a large pair-breaking effect and a significant Tc reduction. The appropriate lengthscale is discussed below.

In the Ho F state, there is a further reduction of Tc (Figs. 4(a) and 4(b)) which can be explained by the enhanced mean exchange field. This behavior is directly analogous to the reduction of Tc observed in F/S/F and F/F/S heterostructures when the F layers were parallel rather than antiparallel.4–9 The advantage of our architecture is that we are able to control Tc by changing the magnetic state of a single layer. Moreover, the magnitude of ΔTc (130 mK) is almost an order of magnitude larger than those reported in spin valve experiments. In those heterostructures, the mean exchange field decays over a relatively long distance before it acts on Cooper pairs, particularly for F/F/S heterostructures.

However, ΔTc is an order of magnitude smaller than the initial suppression by epitaxial Ho because the vertical averaging length (ξHo) is more critical than the horizontal averaging length (ξNb) for all-epitaxial Nb/Ho bilayers. From the measured resistivity of a 100 nm epitaxial Ho film (95 μΩ cm), we can calculate the electron mean free path (lHo) and coherence length (ξHo) as 0.7 and 5 nm, respectively (assuming the Fermi velocity as 1.6 × 106 m/s23). As λ = 3.4 nm ∼ ξHo, the exchange field acting on Cooper pairs in AF Ho does not average out, but giving a smaller Tc reduction compared with the F state.

Apart from the singlet exchange interaction considered above, triplet pair generation may also account for the observed results. It has been experimentally24–26 and theoretically27–30 proved that Ho can convert the singlet state into spin-aligned triplet state. This conversion can open a proximity channel for singlet Cooper pairs into linear ferromagnets, resulting in Tc reduction.31–33 Because the magnetic inhomogeneity is enhanced in epitaxial Ho as opposed to non-epitaxial Ho, the conversion by epitaxial Ho can become effective and the effective conversion could cause the suppression of Tc observed but the enhanced Tc suppression for the F state cannot be explained on this basis.

We therefore conclude that the direct manipulation by gradually changing the magnetic states of a single epitaxial Ho layer controls the Tc of Nb thin films via a proximity effect controlled by the mean exchange energy of the Ho.

This work is partially supported by ERC AiG “Superspin.” Yuanzhou Gu acknowledges financial support from King's College, Cambridge.

1.
M.
Eschrig
,
Phys. Today
64
(
1
),
43
(
2011
).
2.
S.
Oh
,
D.
Youm
, and
M. R.
Beasley
,
Appl. Phys. Lett.
71
,
2376
(
1997
).
3.
L.
Tagirov
,
Phys. Rev. Lett.
83
,
2058
(
1999
).
4.
J. Y.
Gu
,
C.-Y.
You
,
J. S.
Jiang
,
J.
Pearson
,
Y. B.
Bazaliy
, and
S. D.
Bader
,
Phys. Rev. Lett.
89
,
267001
(
2002
).
5.
A.
Potenza
and
C. H.
Marrows
,
Phys. Rev. B
71
,
180503
(
2005
).
6.
I. C.
Moraru
,
W. P.
Pratt
 Jr.
, and
N. O.
Birge
,
Phys. Rev. Lett.
96
,
037004
(
2006
).
7.
G.-X.
Miao
,
A.
Ramos
, and
J. S.
Moodera
,
Phys. Rev. Lett.
101
,
137001
(
2008
).
8.
G.
Nowak
,
H.
Zabel
,
K.
Westerholt
,
I.
Garifullin
,
M.
Marcellini
,
A.
Liebig
, and
B.
Hjörvarsson
,
Phys. Rev. B
78
,
134520
(
2008
).
9.
P. V.
Leksin
,
N. N.
Garif'yanov
,
I. A.
Garifullin
,
J.
Schumann
,
H.
Vinzelberg
,
V.
Kataev
,
R.
Klingeler
,
O. G.
Schmidt
, and
B.
Büchner
,
Appl. Phys. Lett.
97
,
102505
(
2010
).
10.
P. V.
Leksin
,
N. N.
Garif'yanov
,
I. A.
Garifullin
,
J.
Schumann
,
V.
Kataev
,
O. G.
Schmidt
, and
B.
Büchner
,
Phys. Rev. Lett.
106
,
067005
(
2011
).
11.
P. V.
Leksin
,
N. N.
Garif'yanov
,
I. A.
Garifullin
,
J.
Schumann
,
V.
Kataev
,
O. G.
Schmidt
, and
B.
Büchner
,
Phys. Rev. B
85
,
024502
(
2012
).
12.
P. V.
Leksin
,
A. A.
Kamashev
,
N. N.
Garif'yanov
,
I. A.
Garifullin
,
Y. V.
Fominov
,
J.
Schumann
,
C.
Hess
,
V.
Kataev
, and
B.
Büchner
,
JETP Lett.
97
,
478
(
2013
).
13.
B. L.
Rhodes
,
S.
Legvold
, and
F. H.
Spedding
,
Phys. Rev.
109
,
1547
(
1958
).
14.
D. L.
Strandburg
,
S.
Legvold
, and
F. H.
Spedding
,
Phys. Rev.
127
,
2046
(
1962
).
15.
R. S.
Safrata
,
T. R.
Fisher
, and
E. G.
Shelley
,
J. Appl. Phys.
37
,
4869
(
1966
).
16.
W. C.
Koehler
,
J. W.
Cable
,
M. K.
Wilkinson
, and
E. O.
Wollan
,
Phys. Rev.
151
,
414
(
1966
).
17.
W. C.
Koehler
,
J. W.
Cable
,
H. R.
Child
,
M. K.
Wilkinson
, and
E. O.
Wollan
,
Phys. Rev.
158
,
450
(
1967
).
18.
J. D. S.
Witt
,
T. P. A.
Hase
,
R.
Fan
,
T. R.
Charlton
,
S.
Langridge
, and
M. G.
Blamire
,
J. Phys. Condens. Matter
23
,
416006
(
2011
).
19.
J.
Kwo
,
M.
Hong
, and
S.
Nakahara
,
Appl. Phys. Lett.
49
,
319
(
1986
).
20.
J. D. S.
Witt
,
J. W. A.
Robinson
, and
M. G.
Blamire
,
Phys. Rev. B
85
,
184526
(
2012
).
21.
J.
Stöhr
and
H.
Siegmann
,
Magnetism: From Fundamentals to Nanoscale Dynamics
(
Springer-Verlag
,
Berlin
,
2006
).
22.
F.
Chiodi
,
J. D. S.
Witt
,
R. G. J.
Smits
,
L.
Qu
,
G. B.
Halász
,
C.-T.
Wu
,
O. T.
Valls
,
K.
Halterman
,
J. W. A.
Robinson
, and
M. G.
Blamire
,
Europhys. Lett.
101
,
37002
(
2013
).
23.
G.
Halász
,
J.
Robinson
,
J.
Annett
, and
M.
Blamire
,
Phys. Rev. B
79
,
224505
(
2009
).
24.
J. W. A.
Robinson
,
J. D. S.
Witt
, and
M. G.
Blamire
,
Science
329
,
59
(
2010
).
25.
I.
Sosnin
,
H.
Cho
,
V. T.
Petrashov
, and
A. F.
Volkov
,
Phys. Rev. Lett.
96
,
157002
(
2006
).
26.
I. T. M.
Usman
,
K. A.
Yates
,
J. D.
Moore
,
K.
Morrison
,
V. K.
Pecharsky
,
K. A.
Gschneidner
,
T.
Verhagen
,
J.
Aarts
,
V. I.
Zverev
,
J. W. A.
Robinson
,
J. D. S.
Witt
,
M. G.
Blamire
, and
L. F.
Cohen
,
Phys. Rev. B
83
,
144518
(
2011
).
27.
M.
Alidoust
and
J.
Linder
,
Phys. Rev. B
82
,
224504
(
2010
).
28.
G. B.
Halász
,
M. G.
Blamire
, and
J. W. A.
Robinson
,
Phys. Rev. B
84
,
024517
(
2011
).
29.
L.-J.
Jin
,
Y.
Wang
,
L.
Wen
,
G.-Q.
Zha
, and
S.-P.
Zhou
,
Phys. Lett. A
376
,
2435
(
2012
).
30.
C.-T.
Wu
,
O. T.
Valls
, and
K.
Halterman
,
Phys. Rev. B
86
,
184517
(
2012
).
31.
Y.
Fominov
and
A.
Golubov
,
JETP Lett.
91
,
308
(
2010
)
Y.
Fominov
and
A.
Golubov
, [
Pis'ma ZhETF
91
,
329
(
2010
)].
32.
P. V.
Leksin
,
N. N.
Garif'yanov
,
I. A.
Garifullin
,
Y. V.
Fominov
,
J.
Schumann
,
Y.
Krupskaya
,
V.
Kataev
,
O. G.
Schmidt
, and
B.
Büchner
,
Phys. Rev. Lett.
109
,
057005
(
2012
).
33.
N.
Banerjee
,
C. B.
Smiet
,
R. G. J.
Smits
,
A.
Ozaeta
,
F. S.
Bergeret
,
M. G.
Blamire
, and
J. W. A.
Robinson
,
Nat. Commun.
5
,
3048
(
2014
).